Computer Aided Surgery 5:156 –165 (2000)

Biomedical Paper
A Simulator for Maxillofacial Surgery Integrating 3D
Cephalometry and Orthodontia
G. Bettega, M.D., Ph.D., Y. Payan, Ph.D.,
B. Mollard, M.Sc., A. Boyer, M.Sc., B. Raphaël, M.D., Ph.D.,
and S. Lavallée, Ph.D.
Service de Chirurgie Plastique et Maxillo-Faciale, Centre Hospitalier Universitaire de Grenoble (G.B.,
B.R.), Laboratoire TIMC/IMAG, Université Joseph Fourier de Grenoble (Y.P., B.M.), and PRAXIM,
La Tronche (A.B., S.L.), France

ABSTRACT Objectives: This paper presents a new simulator for maxillofacial surgery that gathers
the dental and maxillofacial analyses together into a single computer-assisted procedure. The idea is
to first propose a repositioning of the maxilla via the introduction of 3D cephalometry applied to a
3D virtual model of the patient’s skull. Orthodontic data are then integrated into this model, using
optical measurements of plaster casts of the teeth.
Materials and Methods: The feasibility of the maxillofacial demonstrator was first evaluated
on a dry skull. To simulate malformations (and thus simulate a “real” patient), the skull was modified
and manually cut by the surgeon to generate a given maxillofacial malformation (with asymmetries
in the sagittal, frontal, and axial planes).
Results: The validation of our simulator consisted of evaluating its ability to propose a bone
repositioning diagnosis that would restore the skull to its original configuration. An initial qualitative
validation is provided in this paper, with a 1.5-mm error in the repositioning diagnosis.
Conclusions: These results mainly validate the concept of a maxillofacial numerical simulator
that integrates 3D cephalometry and guarantees a correct dental occlusion. Comp Aid Surg 5:156 –165
(2000). ©2000 Wiley-Liss, Inc.

Key words: maxillofacial surgery, simulator, orthodontia, cephalometry

INTRODUCTION
Planning craniofacial surgical procedures, particu-
larly orthognathic surgery, requires the integration
of multiple and complex data gathered from differ-
ent sources: clinical examination (anthropometry),
orthodontic (dental models), radiological (cepha-
lometry), and intra-operative data (constraints and
position information). This heterogeneity makes
the therapeutic decision difficult, particularly in
asymmetrical dysmorphoses. For this reason, sev-
eral types of three-dimensional (3D) surgical anal-
ysis, simulation software, and methods have been
developed.3,7,9 –11,13–15,17–20 The pioneers in this
field were Marsh and Vannier.13,18,19 According to
Cutting,7 a surgical simulation program must be
built with three functions. First, one must be able to
cut a model of the skull in ways that reflect actual
surgical procedures. Second, mobilization of the
bone segments with six degrees of freedom must be

Received January 12, 2000; accepted March 3, 2000.

Address correspondence/reprint requests to: Yohan Payan, Ph.D., Laboratoire TIMC, équipe GMCAO – Institut Albert
Bonniot, Faculté de Médecine, 38706 La Tronche Cedex, France; Telephone: 33 4-76-54-95-22; Fax: 33 4-76-54-95-55;
E-mail: yohan.payan@imag.fr.
©2000 Wiley-Liss, Inc.
 Bettega et al.: Simulator for Maxillofacial Surgery 157

Fig. 1. Delaire cephalometry on sagittal radiographic tracings: computation of maxilla and mandible repositioning achieved
with tracing paper.

possible. The third function is to create a 3D ceph- without degradation of the database, and theoreti-
alometric analysis. To these functions, one must cally allow combination of osseous and soft-tissue
add the necessity of being able to adapt to the simulation. The digital data format facilitates quan-
limitations imposed by the anatomical or physio- titative analysis of the simulation and outcome, but
logical characteristics of the area (thereby preserv- it is very difficult to define the dental occlusion
ing vessels, nerves, etc.). It is also important to with sufficient accuracy. From this point of view,
integrate a soft-tissue simulation in this bone anal- stereolithographic models are more concrete for the
ysis. surgeon;6 the occlusal problem can be solved, and
Another important challenge is transferring implants can be prepared prior to surgery. How-
these 3D data to the operating room in order to ever, the manipulation is destructive, and the cost
simplify the surgical procedure with the aid of the and fabrication time of the model are shortcomings
computer.4,5 Even though the technology is rapidly of this procedure.
improving, the simulations proposed are still rudi- This paper deals with a 3D cephalometric
mentary17 and do not take into account previous 3D analysis system and a surgical simulator for orthog-
cephalometric analyses. Three-dimensional cepha- nathic surgery that integrates the advantages of
lometric analysis is a real problem and very few
relevant publications are available. The major dif-
ficulties are the large volume of data that has to be
processed by the computer and the lack of cranio-
facial normative 3D data. Altobelli1 has discussed
the use of anthropometric data or the extrapolation
of two-dimensional (2D) data. Marsh13 considers
that this extrapolation is adequate in cases of sym-
metrical dysmorphosis, but cannot be applied to
craniofacial problems or asymmetrical abnormali-
ties.
Surgical simulation is usually performed in
two environments: digital graphics workstations
and solid life-size skull facsimiles. Digital graphics Fig. 2. A “standard” dry skull (left) manually cut to
workstations allow multiple simulated operations simulate malformations (right).
158 Bettega et al.: Simulator for Maxillofacial Surgery
Fig. 3. Software interface: horizontal CT slices (top) and
the corresponding reconstructed 3D model of the patient’s
skull (bottom).
both environments (graphic and facsimiles). The
simulator is based on the integration of dental mod-
els and 3D cephalometry.
OBJECTIVES
State of the Art
Orthognathic surgery deals with face dysmorphosis
arising from congenital malformations or acci-
dents.16 For example, in the case of mandibular
prognathism (a dentofacial deformity of the lower
third of the face resulting from excess mandibular
growth), orthognathic surgical treatment is required
to correct the occlusion (dental position) by means
of an osteotomy of the mandible.2 For this treat-
ment, and for many other orthognathic surgical
procedures, surgeons usually start from (1) plaster
casts of the teeth and (2) sagittal, frontal, and/or
axial 2D radiographs of the patient’s head.
Plaster casts of the teeth are used to plan the
osteotomy phase: both casts (of the mandible and
maxilla) are manually cut to simulate (1) a correct
positioning of the maxilla in relation to some spe-
cific facial anatomical landmarks (by means of a
facial bow study), and (2) a correct positioning of
the mandible in relation to the maxilla that guaran-
tees normal dental occlusion. During this cutting
procedure, resin splints (called intercuspidation
splints) are built from the plaster casts, providing
dental occlusion prints for the initial (actual maxilla
and mandible), intermediary (actual mandible and
cut maxilla) and final (cut maxilla and mandible)
plaster cast positions. These splints are used during
surgery as references for maxillary and mandibular
osteotomies.
In parallel to this dental planning, surgeons
can make a simplified 2D Delaire cephalometry,8
i.e., compute from sagittal and/or facial radio-
graphic tracings the desired displacements of the
maxilla and the mandible. This is achieved by first
placing specific anatomical landmarks onto the ra-
diography, then, using tracing paper, suitable dis-
placements of mandibular and maxillary landmarks
in relation to the rest of the skull are measured
manually (Fig. 1). This cephalometric diagnosis is
then compared to the displacements provided by
the orthodontic facial bow study.
As may be deduced from the above, two
parallel procedures are required for the planning of
orthognathic surgery: dental analysis and maxillo-
facial analysis, both working on different princi-
ples. Moreover, the decision phase only occurs at
the end of each procedure, which means that the
undertaking has been a waste of time if the two
plans are not compatible and the procedures have to
be run again.
The aim of the work presented in this paper is
to combine the two surgical planning procedures
into a single computer-assisted procedure that in-
tegrates information from surgeons about reposi-
tioning of the bone structures, and information
from orthodontists about optimal dental occlusion.
The idea is first to use orthopedic knowledge to
propose a repositioning of the maxilla via the in-
troduction of a 3D cephalometry applied to a 3D
virtual model of the patient skull. Then, orthodontic
data obtained by measurements of plaster casts of
the teeth are integrated into this model using a 3D
localizing system. For this approach, only the final
desired position of the mandible, in relation to the
 Bettega et al.: Simulator for Maxillofacial Surgery
Fig. 4. 3D extrapolation (right) of the simplified Delaire analysis (left).
maxilla, is taken into account. Therefore, no cast-
cutting phase is required, which makes the proce-
dure easier.
MATERIALS AND METHODS
Choice of Patient
The feasibility of our demonstrator was first eval-
uated on a dry skull. This skull was a “standard”
one without any noticeable maxillofacial dysmor-
phosis (Fig. 2, left). This choice was motivated by
our wish to be able to quantify the repositioning
diagnosis proposed by the simulator, which neces-
sitates having knowledge of the “normal” maxillary
and mandibular positions. To simulate malforma-
tions (and thus a “real” patient), the skull was
modified and manually cut by the surgeon (Fig. 2,
right) to generate a given maxillofacial dysmorpho-
sis (with asymmetries in the sagittal, frontal, and
axial planes). Before this cutting phase, two paral-
lel tubes were fixed between the forehead and the
mandible.
As can be seen in Figure 2 (right panel), each
tube had to be cut into three parts in order to allow
159
the manual bone-structure cutting phase. The vali-
dation of our simulator would thus consist of eval-
uating its ability to propose a repositioning diagno-
sis that re-aligned each part of the two tubes, as in
the original skull configuration.
Data Acquisition and 3D Reconstruction of
the Patient’s Skull
Horizontal Computer Tomography (CT) slices
were collected for the whole skull (helical scan
with a 3-mm pitch and slices reconstructed every
1.5-mm). The Marching Cubes algorithm12 has
been implemented to reconstruct the skull from CT
slices. Before running this reconstruction process,
tools (erasers) can be used to clean specific slices,
and a threshold value for the reconstructed isosur-
face must be chosen (the top panel in Figure 3
shows a snapshot of the PC platform software).
Then, the process automatically builds the virtual
3D model (Figure 3, lower panel).
3D Cephalometry
The third dimension brings to cephalometric anal-
ysis the advantage of taking into account the data
160 Bettega et al.: Simulator for Maxillofacial Surgery

Fig. 5. Positioning of the anatomical landmarks (upper panels) and the corresponding 3D cephalometric analysis (lower
panel).

provided by frontal, sagittal, and axial studies in a
single step. It allows the integration of the problems
of facial asymmetry and occlusal plane horizontal-
ity into the profile analysis. Apart from the imple-
mentation, the main problem in 3D cephalometry is
the standardization and reference to the norms that
exist in 2D cephalometries. Instead of creating a
new 3D analysis, the idea was to transpose the data
from 2D cephalometry in the third dimension. Our
approach consists of a 3D extrapolation of the
simplified Delaire analysis and is illustrated in Fig-
ure 4.
 Bettega et al.: Simulator for Maxillofacial Surgery
Fig. 6. A cube for simulating a virtual osteotomy that separates the maxilla/mandible block from the rest of the skull.
This analysis is adapted to the third dimen-
sion so that the reference standards existing in the
sagittal plane are respected. The norms in the other
dimensions are theoretically easy to define: it is
simply a matter of respecting the horizontality in
the frontal plane and the symmetry in relation to the
sagittal median plane.
The surgeon is therefore asked to manually
position each of the points listed in Figure 4 onto
the virtual model of the patient’s skull (Fig. 5, top
panels). Starting from these cephalometric points,
an automatic analysis procedure provides specific
lines and planes (Fig. 5, lower panel) which will be
used for the determination of a repositioning diag-
nosis.
Before this diagnosis is made, pixels of the
3D model belonging to the maxilla/mandible block
must be labeled, as the repositioning diagnosis will
be applied to these points. For this step, a virtual
osteotomy is manually simulated that separates the
skull model into two groups of points (Fig. 6).
As shown in Figure 6, the virtual osteotomy is
161
performed using a parallelepiped cutting pattern
that is interactively placed on the skull model and
dimensionally adjusted with the manipulation tools
provided by the software. These tools are sufficient
to obtain a realistic model of surgical cutting.
RESULTS
Maxilla Repositioning Diagnosis
Maxilla repositioning is totally driven by the ceph-
alometric analysis, according to the following three
constraints (see Figure 4 for the names of planes
and points):
i) The NP point is moved to fit the theoretical
NP point position computed from cephalo-
metric analysis (onto the intersection be-
tween the CF1 plane and the sagittal median
(SM) plane).
ii) The CF7 plane is moved to fit the theoretical
CF7 plane.
iii) A given point chosen at the intersection
162 Bettega et al.: Simulator for Maxillofacial Surgery
Fig. 7. A diagnosis for the repositioning of the maxilla/
mandible block (lower panels), automatically obtained from
the 3D cephalometric analysis.
between the two patient incisors is moved
so as to be projected onto the sagittal me-
dian plane.
Following those constraints, a global dis-
placement of the maxillary structure is computed in
terms of translation and rotation. Figure 7 (lower
panels) plots the corresponding repositioning diag-
nosis (NB: mandible position, in relation to the
maxilla, remains constant during this operation).
Figure 8. Intercuspidation splints inserted into plaster casts of teeth, localized by the means of optical rigid bodies. Left:
actual position; right: desired dental occlusion.
Mandible Repositioning Diagnosis
As the mandible repositioning has to integrate den-
tal occlusion constraints, it was decided to let it be
totally driven by dental diagnosis. Cephalometric
points and planes resulting from our analysis were
here only used to determine the real occlusion
plane (CF7) between the maxilla and mandible (in
order to label points according to whether they
belonged to the maxilla or to the mandible).
As in standard treatments, orthodontic diag-
nosis was carried out on dental plaster casts. The
only concern was the position of the mandible in
relation to the maxilla, in terms of optimal dental
occlusion. In contrast to standard orthognathic pro-
cedures, however, plaster casts of the teeth did not
have to be cut, as there was no need to be con-
cerned about maxilla positioning relative to the rest
of the skull.
The initial and final splints were used for the
integration of the orthodontic diagnosis into the
virtual model of the patient’s skull (the intermedi-
ary splint is thus removed from the procedure).
Inserted into the plaster casts of the teeth, these
splints respectively replicate the current and de-
sired positions of the mandible in relation to the
maxilla. A 3D optical localizer (Optotrak™, North-
ern Digital Inc.) is used to quantitatively measure
the corresponding displacement from the current
position (Fig. 8, left panel) to the desired position
(Fig. 8, right panel).
This measurement consists of a global trans-
formation matrix, corresponding to the desired
translation and rotation that has to be applied to the
 Bettega et al.: Simulator for Maxillofacial Surgery
Fig. 9. Initial intercuspidation splint with two aluminum
tubes.
mandible. As this transformation is expressed in the
localizer referential, it has to be transferred into the
CT scans’ referential, i.e., into the virtual model
space. To do this, an object visible in both modal-
ities (localizer space and CT space) has been intro-
duced into the procedure. This object comprises a
pair of aluminum tubes fixed onto the initial inter-
cuspidation splint (Fig. 9).
As these tubes are made from aluminum, they
can be detected on CT scans if they are inserted
Fig. 10. Measurements of the aluminum tubes’ axes in the localizer referential (left) and in the CT scans referential (right).
163
inside the patient’s mouth during the CT recordings
(Fig. 10, right panel). Moreover, they can also be
detected and located in the optical localizer refer-
ential (Fig. 10, left panel). Therefore, a simple
matching algorithm enables us to compute the
transformation from one referential to the other,
and thus transpose the mandible correction into the
virtual model space (Fig. 11, lower panel).
DISCUSSION
The repositioning diagnosis simulation presented in
Figure 11 validates the feasibility of our simulator,
as each part of the two tubes is qualitatively re-
aligned with the other ones. These tubes show a
maximal deviation of 2 degrees between their axes,
which roughly corresponds to a 1.5-mm error in the
repositioning procedure. These results mainly val-
idate the concept of a maxillofacial numerical sim-
ulator that (1) integrates 3D cephalometry, (2)
guarantees a correct dental occlusion, and (3) pro-
poses a semi-automatic diagnosis for maxilla and
mandible repositioning. However, the errors ob-
tained are not completely satisfying, as the aim for
orthognathic surgery would be to have a precision
below one millimeter. The next stage of this work
will thus consist of a quantitative evaluation of
each point of the global procedure: precision of the
computer-assisted 3D cephalometry, precision of
the dental occlusion measurement, and repeatabil-
164 Bettega et al.: Simulator for Maxillofacial Surgery
Fig. 11. Integration of the orthodontic mandible reposi-
tioning diagnosis (lower panels) into the virtual model of
the patient’s skull.
ity of the process for the same patient and/or for
different kinds of pathology. Moreover, a compar-
ison of the simulator diagnosis with osteotomies
provided by surgeons using classical procedures
without computer-assisted techniques would be an-
other important evaluation factor.
CONCLUSION
In this paper we have presented an initial evalua-
tion of the feasibility of computer-assisted tech-
niques for maxillofacial surgery, integrating 3D
cephalometry and orthodontic information. The
first tests have been carried out on a dry skull that
was manually cut to simulate a pathological patient.
The repositioning diagnosis proposed by the max-
illofacial simulator was evaluated through re-align-
ment of tubes fixed on the cadaver skull before the
simulation of the patient dysmorphosis. A first clue
to the efficiency of the simulator was provided and
discussed. Clinical tests must now be carried out on
real patients, and with different types of patholo-
gies. Then, the next phase of this work will consist
of transferring the simulated repositioning diagno-
sis into the operating room.
ACKNOWLEDGMENTS
This work was financially supported by the Euro-
pean project IGOS HC1026HC.
REFERENCES
1. Altobelli DE, Kikinis R, Mulliken JB, Cline H,
Lorensen W, Jolesz F. Computer-assisted three-di-
mensional planning in craniofacial surgery. Plastic
Reconstruc Surg 1993;92:576 –585, discussion 586 –
587.
2. Bell H. Modern practice in orthognathic and recon-
structive surgery. Vol 3. Philadelphia: WB Saunders
Company; 1992.
3. Bettega G, Chenin M, Sadek H, Cinquin P, Lebeau J,
Coulomb M, Raphaël B. Three dimensional fetal
cephalometry. Cleft Palate Craniofac J 1996;33:463–
467.
4. Bettega G, Dessenne V, Cinquin P, Raphaël B. Com-
puter assisted mandibular condyle positioning in or-
thognatic surgery. J Oral Maxillofac Surg 1996;54:
553–558.
5. Bettega G, Cinquin P, Lavallée S, Raphaël B. Disposi-
tif de détermination d’un déplacement entre des posi-
tions relatives de deux moulages dentaires et système
de simulation d’intervention en chirurgie orthog-
nathique. Patent FR 9616066; 1996.
6. Bill JS, Reuther JF, Dittmann W, Kübler N, Meier JL,
Pistner H, Wittenberg G. Stereolithography in oral and
maxillofacial operation planning. Int J Oral Maxillo-
fac Surg 1995;24:98 –101.
7. Cutting C, Bookstein FL, Grayson B, Fellingham L,
McCarthy JG. Three-dimensional computer-assisted
design of craniofacial surgical procedures: optimiza-
tion and interaction with cephalometric and CT-based
models. Plastic Reconstruc Surg 1986;77:877– 885.
8. Delaire J. L’analyse architecturale et structurale
cranio-faciale (de profil). Principes théoriques.
Quelques exemples d’emploi en chirurgie maxillo-
faciale. Rev Stomatol Chirurgie Maxillo Faciale 1978;
79:1–33.
9. Keeve E, Girod S, Girod B. Computer-aided cranio-
facial surgery. In: Lemke HU, Vannier MW, Inamura
K, Farman AG, editors: Computer Assisted Radiol-
ogy: Proceedings of the International Symposium on
Computer and Communication Systems for Image
Guided Diagnosis and Therapy (CAR’96). Amster-
dam: Elsevier; 1996. p 757–763.
10. Keeve E, Girod S, Kikinis R, Girod B. Deformable
modeling of facial tissue for craniofacial surgery sim-
ulation. Comp Aid Surg 1998;3:228 –238.
11. Lo LJ, Marsh JL, Vannier MW, Patel VV. Craniofa-
cial computer assisted surgical planning. Clin Plastic
Surg 1994;21:501–516.
12. Lorensen WE, Cline HE. Marching Cubes: A high
resolution 3D surface construction algorithm. ACM
Computer Graphics 1987;21:163–169.
13. Marsh JL, Vannier MW, Bresina S, Hemmer KM.
Applications of computer graphics in craniofacial sur-
gery. Clin Plast Surg 1986;13:441– 448.
14. Mollard B, Lavallée S, Bettega G. Computer assisted
orthognathic surgery. In: Wells WM, Colchester A,
Delp S, editors: Proceedings of the First International

Conference on Medical Image Computing and Com-
puter-Assisted Intervention (MICCAI’98), Cam-
bridge, MA, October 1998. Lecture Notes in Com-
puter Science 1496. Berlin, Heidelberg, New York:
Springer-Verlag; 1998. p 21–28.
15. Schutyser F, Van Cleynenbreugel J, Schoenaers J,
Marchal G, Suetens P. A simulation environment for
maxillofacial surgery including soft tissue implica-
tions. In: Taylor C, Colchester A, editors: Proceedings
of Second International Conference on Medical Image
Computing and Computer-Assisted Intervention
(MICCAI’99), Cambridge, UK, September 1999. Ber-
lin, Heidelberg, New York: Springer-Verlag; 1999. p
1210 –1217.
16. Stricker M, Van Der Meulen J, Raphaël B, Mazzola R.
Craniofacial Malformations. Churchill Livingstone,
Longman Group; 1990.
Bettega et al.: Simulator for Maxillofacial Surgery 165
17. Udupa JK, Odhner D. Fast visualization, manipula-
tion, and analysis of binary volumetric objects. IEEE
Computer Graphics & Applications 1991. p 53– 62.
18. Vannier MW, Marsh JL, Tsiaras A. Craniofacial sur-
gical planning and evaluation with computers. In:
Taylor R, Lavallée S, Burdea G, Mösges R, editors:
Computer-Integrated Surgery. Cambridge, MA: MIT
Press; 1996. p 673– 677.
19. Vannier MW, Marsh JL, Warren JO. Three-dimen-
sional computer graphics for craniofacial surgical
planning and evaluation. Computer Graphics 1983;17:
263–274.
20. Waters K. Synthetic muscular contraction on facial
tissue derived from computerized tomography data.
In: Taylor R, Lavallée S, Burdea G, Mösges R, edi-
tors: Computer-Integrated Surgery. Cambridge, MA:
MIT Press; 1996. p 191–199.